identify and add missing stations from train routes

2 files changed, 90 insertions, 22 deletions

diff --git a/share/find-missing b/share/find-missing
index 909c203..e69a7a1 100755
--- a/share/find-missing
+++ b/share/find-missing
@@ -1,13 +1,59 @@
-#!/bin/zsh
-
-: > missing.txt
-
-./xml2json | fgrep missing | shuf | while read m eva x ; do
-	if [[ $x != *,* ]] && db-iris -x $eva &> /dev/null; then
-		echo "\n$m $eva $x"
-		echo $m $eva $x >> missing.txt
-	else
-		echo -n .
-	fi
-	sleep .2
-done
+#!/usr/bin/env perl
+
+use strict;
+use warnings;
+use 5.010;
+
+use Cache::File;
+use List::Util qw(uniq);
+use Travel::Status::DE::IRIS;
+use Travel::Status::DE::IRIS::Stations;
+
+my ($station) = @ARGV;
+
+my $cache_path          = $ENV{XDG_CACHE_HOME} // "$ENV{HOME}/.cache";
+my $schedule_cache_path = "${cache_path}/db-iris-schedule";
+my $realtime_cache_path = "${cache_path}/db-iris-realtime";
+my ( $schedule_cache, $realtime_cache );
+
+eval {
+	use Cache::File;
+	$schedule_cache = Cache::File->new(
+		cache_root      => $schedule_cache_path,
+		default_expires => '6 hours',
+		lock_level      => Cache::File::LOCK_LOCAL(),
+	);
+	$realtime_cache = Cache::File->new(
+		cache_root      => $realtime_cache_path,
+		default_expires => '180 seconds',
+		lock_level      => Cache::File::LOCK_LOCAL(),
+	);
+};
+if ($@) {
+	$schedule_cache = undef;
+	$realtime_cache = undef;
+}
+
+my $status = Travel::Status::DE::IRIS->new(
+	station        => $station,
+	main_cache     => $schedule_cache,
+	realtime_cache => $realtime_cache,
+	with_related   => 1,
+);
+
+my @missing;
+
+for my $result ( $status->results ) {
+	for my $name ( $result->route_pre, $result->route_post ) {
+		my @matches
+		  = Travel::Status::DE::IRIS::Stations::get_station_by_name($name);
+		if ( @matches != 1 ) {
+			push( @missing, $name );
+		}
+		elsif ( $matches[0][1] ne $name ) {
+			push( @missing, $name );
+		}
+	}
+}
+
+say join( "\n", uniq @missing );
diff --git a/share/xml2json b/share/xml2json
index 951d578..84a1594 100755
--- a/share/xml2json
+++ b/share/xml2json
@@ -6,6 +6,7 @@ use 5.010;
 
 use File::Slurp qw(read_file write_file);
 use JSON;
+use List::Util qw(any);
 use XML::LibXML;
 
 my $json_str = read_file('stations.json');
@@ -15,6 +16,15 @@ my $stations = JSON->new->utf8->decode($json_str);
 my $xml_str = read_file('stations.xml');
 my $tree    = XML::LibXML->load_xml( string => $xml_str );
 
+my @missing;
+
+if ( -e "missing.txt" ) {
+	for my $line ( read_file("missing.txt") ) {
+		chomp $line;
+		push( @missing, $line );
+	}
+}
+
 for my $station ( $tree->findnodes('//station') ) {
 	my $name  = $station->getAttribute('name');
 	my $eva   = $station->getAttribute('eva');
@@ -44,25 +54,37 @@ for my $station ( $tree->findnodes('//station') ) {
 				$eva, $j_name, $name, $j_ds100, $ds100 );
 			last;
 		}
-		elsif ( $j_name eq $name and $j_ds100 ne $ds100 and $is_db ) {
+		elsif ( $j_name eq $name
+			and $j_ds100 ne $ds100
+			and $is_db
+			and $ds100 !~ m{ ^ PQ }x )
+		{
 			printf( "%30s has been recoded: %8s -> %8s\n",
 				$name, $j_ds100, $ds100 );
 			last;
 		}
-		elsif ( $j_name eq $name and $j_eva != $eva and $is_db ) {
+		elsif ( $j_name eq $name
+			and $j_eva != $eva
+			and $is_db
+			and $ds100 !~ m{ ^ PQ }x )
+		{
 			printf( "%30s has been recoded: %d -> %d\n", $name, $j_eva, $eva );
 			last;
 		}
 	}
 
-	if (
-		not $found
-		and not( $ds100 =~ m{ ^ ( [OPXZ] | D[0-9] ) }x
-			or $name =~ m{ ^ Bahnhof, }x
-			or $name =~ m{ \(Gr\) $ }x )
-	  )
+	if ( not $found
+		and any { $_ eq $name } @missing )
 	{
-		#say "missing $eva  $ds100  \"$name\"";
+		say "missing $eva  $ds100  \"$name\"";
+		push(
+			@{$stations},
+			{
+				name  => $name,
+				ds100 => $ds100,
+				eva   => $eva,
+			}
+		);
 	}
 }